db 0 is always emitted at the end of a script

[ISSOtm]

I'm not sure why it's always a zero. It seems to be intended as an implicit trailing return, but then I'm not sure why it's not emitted as such? (If return is not zero, that's another problem.)

Additionally, it'd be nice if it could be removed for scripts that don't need it. IMO it shouldn't be a default behaviour, in fact.

[evie-calico]

I'm not comfortable making no return the default, but a command line flag to toggle it would be nice and the 0 should obviously be corrected to a return byte.

[ISSOtm]

Actually, I think such configuration should belong to the file itself, as otherwise it means files aren't self-contained—which I think you'll agree it not great.

Perhaps that should be configured in the env block?

[evie-calico]

I guess something like this would work:

env something_that_shouldnt_return {
    use std;
    noreturn;
}

I'm not sure what to call it though; noreturn is concise but not necessarily self-explanatory.

[ISSOtm]

This seems like something that would belong to more "generic" configuration? Maybe some option(noreturn), or option("noreturn") syntax would work well? (I'd suggest porting the pool = syntax to that as well, so that pool is no longer a keyword?)